Tracking infrequent defects - Compounded plastic (poly propylene + filler) injection

[davis007]

I was hired as a quality engineer. My back ground in Chemical Engineering and I have worked in R&D up to this point in my career. Moving into manufacturing has been a shock, but not necessarily a surprise. When I was hired it was made clear that I was to promote 6 sigma and follow the DMAIC process in all my projects. Knowing absolutely nothing about either I have found this forum very helpful.

I have started a project to improve the quality of an self produced component in our final product. The component is a compounded plastic (poly propylene + filler) and is used to injection mold our product. One of the biggest complaints from the injection molding side of the plant has been that they often must make changes to there process to accommodate changes in the compound. Thus the project is to make the compound more consistent.

For project metrics I would like to track how often a change is required at the injection molder. However, the molding side of the facility seems to be unwilling to track this information. Very rarely they give up on a lot of compound and throw it out. They are willing to track this occurrence. Also, the product undergoes a final visual inspection. Defects are reported and can be sorted into categories that are related to the compound or not. So for project metrics I have decided to track lbs of compound scrapped per week, and number of defects found in the finished article related to the compound.

My problem is that these are both relatively rare occurrences. We may go through 120,000lbs/month of compound but scrap only 400-600 lbs. in the whole month, so many weeks have zero compound scrapped even without any improvement project. We produce 1,000,000 parts/week but may only have 700 parts identified with a compound related defect one week and 50 the next.

OK after a long explanation my question. Does it make sense to use these stand in metrics?

[Jim Wynne]

Originally Posted by davis007

… the injection molding side of the …often must make changes to there process to accommodate changes in the compound. Thus the project is to make the compound more consistent.

Who controls the compounding?

Originally Posted by davis007

So for project metrics I have decided to track lbs of compound scrapped per week, and number of defects found in the finished article related to the compound. My problem is that these are both relatively rare occurrences. We may go through 120,000lbs/month of compound but scrap only 400-600 lbs. in the whole month, so many weeks have zero compound scrapped even without any improvement project. We produce 1,000,000 parts/week but may only have 700 parts identified with a compound related defect one week and 50 the next. OK after a long explanation my question. Does it make sense to use these stand in metrics?

First, it’s good to see an instance of a PPM number that actually means something. Maybe what you need to do is get closer to the process that’s causing the problem (compounding) if you’re sure it’s the source of the problem. You’ll need to be able to track compounded lots through manufacturing in order to be able to isolate compounds that work from those that cause problems. From there you’ll have to concentrate on controlling variation in the compounding process. If you have no history to use, you shouldn’t wait until you have some if there’s significant time between occurrences. Go right after the compounding issue and use subsequent production to verify your efforts. Of course, it should go without saying that there should be some cost justification involved; if it costs $10 to solve a $1 problem you should probably look elsewhere for a project.

[davis007]

JSW05:

The compounding process is controlled by the production manager, he also controls the injection molding process.

I spend a lot of time each day working on the compounding process and there are specific improvements that can be made. The issue I am having is that I walked into a situation were most all the problems in injection molding have historically been blamed on "compound variability" and the company believes that this is currently the biggest problem. It has become ingrained into the whole system. After two months here I am convinced that some (many) of these problems are related to other things but am having problems getting others to work with me on that basis. So I thought that I would tackle the problem by working on the "compound variability" issue first. By showing them that this was being worked and some improvements being realized I thought that I could get some support to work on the “other” problems.

My boss the operations manager has pushed to formalize this project into the 6 sigma frame work. Thus the need to establish good project metrics, follow 6 sigma process and be able to demonstrate the improvement.

I guess I feel funny about the metrics we have established because they bounce around so much as it is. Most examples of 6 sigma projects I have found on the web use metrics that occur more frequently, like a 10% defect rate, or goal of meeting orders in 2 weeks vs. 4. When I look a the histograms for the starting metrics (when the are provided) they have few if any values close to 0. For the metrics we have chosen a lot of the data is 0.

[mike101338]

I have a couple questions for you. First, are you sure this is the program you want to attack first? I ask this because of the numbers you've supplied. If your problems are small, it may be easier to attack something with a greater imapct. This would allow you the ability to show everyone the benefits of a six sigma program on a larger scale, then begin working on smaller issues.

The second question relates to the molding side not wanting to track the data you're asking for. Why? Do they not see the benefit? Do they not want others outside the department aware of the problems. Whenever someone doesnt want to track data I always ask myself those questions.

Lastly, product specific in your plant. You say you use a filler added with the poly-propelyne. what type of filler is this? I ask because I am generating tons of scrap plastics and am always looking for additional sources for recycling purposes.

[Jim Wynne]

Originally Posted by davis007

My boss the operations manager has pushed to formalize this project into the 6 sigma frame work.

Then he should be aware that you need data. If you have no data, or feel that the data you do have is inadequate, then you have to wait until there's more data. If your boss thinks that SS methodology requires waiting for data and producing predictable defects instead of just attacking the problem, then your boss is part of the problem. If good production is dependent upon the efficacy of the compounding process, and you think you have excess variation in the compounding process, then you need to control the compounding process.

Originally Posted by davis007

I guess I feel funny about the metrics we have established because they bounce around so much as it is.

That's a sign of excess variation, of course.

Originally Posted by davis007

Most examples of 6 sigma projects I have found on the web use metrics that occur more frequently, like a 10% defect rate, or goal of meeting orders in 2 weeks vs. 4. When I look a the histograms for the starting metrics (when the are provided) they have few if any values close to 0. For the metrics we have chosen a lot of the data is 0.

Don't worry about other projects, especially what you might have seen on the Web. The data is what it is. If some values are zero or close to it, then your question needs to be, "Why aren't all of the values zero or close to it?"

[davis007]

Mike:
Thanks for the input. I agree this in not necessarily the project I would attack first. Mostly because I can not get an objective view of how big the problem is, or is not, largely due to the molding departments refusal to track anything.. Of course this could mean the problem really doe not exist at which point I am trying to fix nothing. What the molding side has everyone else convinced of is that he problem is huge, related to the “compound variability” and is to difficult to measure to try.

Sounds like a perfect storm for failure to me. I am trying to make the best out of the situation in the hopes that a working quality improvement program with some successes will help sway management to change and force others to change. Therefore I am trying to be sure that I do not start out to far on the wrong foot with the metrics. I like JSW05’s suggestion to ask the question why not all values are close to zero.

I tried to institute a program in which the operators and or technicians would report molding problems on a short 3 question form and submit these together with a sample of the material that was running when the problem occurred and a sample that was running when the problem was resolved. The idea being that I could analyze the two samples to start getting an idea of what constituted “good” vs. “bad” material I got one response in two months, and that was because I happened to be standing at the machine when the problem occurred are reminded the tech working on it about the request. We spent ~$5,000 on an large battery of tests we were totally unable to see a difference in the before and after samples.

I guess part of this is a vent so please excuse me if I did not answer your question.

Regarding your last question sorry the composition is a trade secret.

[mike101338]

Does mangement back you? That is another option, although a last resort. Bring in top management and a couple resistive employees to the table. Have the management make it very clear that you are working on their bahlf to better the company, thus they should comply with your requests if they wish to continue being a part of the company. Drastic, but if you're positive the problem exists it may be necessary.

I understand the trade secret. I'm just trying to find additional sources to pawn off mixed, dirty scrap plastics. Interested in purchasing? I'm cheap!

[davis007]

Mike:

Thanks for the advice at this point I am not sure where management is. I know when I interviewed for the position and explained that my background was not quality they really seemed to talk up the 6 sigma methodology etc. Hence my attraction to this forum, to learn. But after being here a few months I am not so sure that it was not just some flash in the pan.

Regarding the use of your scrap. Sorry the "filler" is actually an active component that we have very tight specifications for, at least that was what I was told. When I went to pull COA to put together a time series of the key raw material parameters for the compound. (Hoping to find some trend or correlation with the compound scrapped.) I learned that we have not received a COA for the "filler" in 12 months. So in addition to fighting an internal customer now I must back track all the way to a raw material supplier. I guess I can't complain that there is not enough work to do.